Some cleanup code is not triggered when using LobbyConnection.abort

[Askaholic]

We had some reports of a bug during TMM where a player who was trying to queue by themselves would get the "Only the party owner may enter the party into a queue" error message. It turns out this can be caused by an old party object not getting cleaned up, do to an aborted connection.

The problem is that there is no standard way for a service to handle cleanup for a closed connection. Each service has its own differently named cleanup function which is called by LobbyConnection.on_connection_lost, but only if LobbyConnection.player exists. However LobbyConnection.abort sets this field to None here:

server/server/lobbyconnection.py

Lines 100 to 105 in 55c42f0

	if self.player:
	self._logger.warning(
	"Client %s dropped. %s", self.player.login, logspam
	)
	self.player_service.remove_player(self.player)
	self.player = None

since on_connection_lost is called after the call to abort, this code never executes:

server/server/lobbyconnection.py

Lines 1138 to 1144 in 55c42f0

	if self.player:
	self._logger.debug(
	"Lost lobby connection removing player %s", self.player.id
	)
	await self.ladder_service.on_connection_lost(self.player)
	self.player_service.remove_player(self.player)
	await self.party_service.on_player_disconnected(self.player)

I think we should add a standard on_connection_lost method to the Service base class and call it for all services from either Server or ServerContext when a disconnect happens.